基于多输出t-s模糊模型的转炉煤气柜位预测

《基于多输出t-s模糊模型的转炉煤气柜位预测》由会员上传分享，免费在线阅读，更多相关内容在学术论文-天天文库。

1、基于多输出T-S模糊模型的转炉煤气柜位预测PredictionforConverterGasHolderLevelsbasedonMulti·-outputT-SFuzzyModelAbstractTheconvertergassystemhasthecharacteristicsofmultiple-variable，strongcouplinganddynamicnonlinearity，whichisasystemthatinvolvesmanyproductionandconsumptio

2、nUSerS，delayandlimitedcapacityofthebufferdevice．Thepredictionofconvertergasholderlevelisoneoftheprerequisitesfortheimprovementofconvertergasutilizationefficiencyinironandsteelenterprises．Byachievingdynamicbalanceofbyproductgassystem，theproductioncosti

3、sreducedthusleadstoenterpriseenergysavingandconsumptionreduction．Intheconvertergassystem，thecontrolofgasholderisthebaseofenergybalanceschedulingfortheworkers．IfitsvariationCanbeeffectivelypredictedbyestablishingareliablepredictionmodelofconvertergassy

4、stem，itwillprovidescientificguidanceanddecision．makingbasisfortheefficientuseofconvertergastoachievedynamicbalanceofthegassystem．Giventhatatypicalconvertergassysteminironandsteelenterprisehastwocabinets，andconsideringthecomplexinfluencefactorsofthesys

5、tem，thispaperproposesamulti．-outputT．．SfuzzymodelingmethodbasedonKalmanfilterestimationforpredictingthedualtanklevelofconvertergas．ConsideringtheconvertergassystemdemandofmodernindustrythatthegasisfirstlyconvertedintothetwoholdersandthenitisusedbytheU

6、SerS，thisstudyestablishedafuzzypredictionmodelwithmul邱leoutputs．Theextractionoffuzzyrulesisdividedintothefollowingtwosteps．Firstly,theantecedentparametersareidentifiedbyusingthefuzzyc-meansclusteringalgorithmbasedonthesampledataofhistoricalholdersflow

7、valueandtwoholdersflowdifferencedataclustering．Thentheparametersofmembershipfunctionareobtainedaccordingtotheresultsofclusteringcenterandpartitionmatrix．Secondly,transformthepremiseparameteridentificationintooverdeterminedlinearequations．Amongthem，cal

8、culatingcoefficientofpseudoinversematrixmayproduceabnormalsolutionintheleastsquaresidentificationparameters．Instead，usingthemethodofconstructingthelinearequations，theoptimalsolutionofthelinearequationsisestimatedbyKalmanfilteralgorithm．Thusthe